Analysis for Network Terms

deciding which ergm terms to use based on the data

Marriage / Cohab Network

Propensity to form rels by age-related term

couple options here:

nodefactor(age) - will give you one term for each age, not a good idea over such large age group

nodefactor(agecat) = one term for each agecat - better, but really only good if each age cat has specific rates and not a general trend (if so see below)

nodecov(age or agecat) - one term, assumes some monotonic rate of change between ages or agecats

strike nodefactor(age), too many terms

let’s look at new activity in last year by age and agecat

AgeCat

Age

Age Mixing

options:

nodematch(agecat) absdiff / absdiffby (age / sqrt age?)

AgeCat

Age

Sqrt Age

Age Squared

Existing Relationships

nodefactor(deg.other, base=?)

## Warning: Setting row names on a tibble is deprecated.

Offsets

sex, concurrent

# Other Network

Propensity to form rels by age-related term

couple options here:

nodefactor(age) - will give you one term for each age, not a good idea over such large age group

nodefactor(agecat) = one term for each agecat - better, but really only good if each age cat has specific rates and not a general trend (if so see below)

nodecov(age or agecat) - one term, assumes some monotonic rate of change between ages or agecats

strike nodefactor(age), too many terms

let’s look at new activity in last year by age and agecat

AgeCat

Age

## Age Mixing {.tabset .tabset-pills .tabset-fade} options:

nodematch(agecat) absdiff / absdiffby (age / sqrt age?)

AgeCat

Age

Sqrt Age

Age Squared

## Existing Relationships

nodefactor(deg.other, base=?)

## Warning: Setting row names on a tibble is deprecated.

# Decisions

Marriage/Cohab Network

1. nodecov(age) seems reasonable
2. absdiff(sqrtage)?
3. nodefactor(deg.other, levels=-1)
4. do i need a deg(0)?
5. offsets: sex, concurrent

Other Network

1. nodefactor(agecat, levels:c(1:2)) ? (keep 15:19 and 20:24)
2. absdiff(sqrtage)
3. nodefactor(deg.marcoh, levels=-1)
4. do i need a deg(0)?
5. offsets: sex